Using a mixture model to generate simulated transaction information

Patent metadata

Field	Value
Publication number	US-10692065-B2
Application number	US-201715626795-A
Country	US
Kind code	B2
Filing date	Jun 19, 2017
Priority date	Jun 19, 2017
Publication date	Jun 23, 2020
Grant date	Jun 23, 2020

Abstract

A device may obtain, for a set of transactions, a set of transaction values associated with a particular industry. The device may determine one or more sample statistical distributions for a probabilistic transaction model by using one or more machine learning techniques. The one or more sample statistical distributions may be similar to one or more actual statistical distributions that are associated with the set of transaction values. The device may generate simulated transaction information using the probabilistic transaction model. The device may perform one or more actions after generating the simulated transaction information.

First claim

What is claimed is:

1. A device, comprising:
one or more processors to:
receive, for a set of transactions, real transaction information associated with a particular industry,
the real transaction information including a set of transaction values;

separate the set of transaction values into one or more data streams based on one or more transaction data types,
the set of transaction values being associated with one or more actual statistical distributions,
a first set of transaction values, of the set of transaction values, being associated with transaction amounts,
the first set of transaction values being separated into a first data stream, and

a second set of transaction values, of the set of transaction values, being associated with a debit transaction or a credit transaction,
the second set of transaction values being separated into a second data stream;

train, using one or more machine learning techniques, a model for simulating the real transaction information;
generate simulated transaction information by using random numbers as input for the model;
process the simulated transaction information to detect an anomaly relating to the real transaction information after generating the simulated transaction information;
cause at least one of a training procedure or a testing procedure to be executed based on detecting the anomaly; and
provide a recommendation to an entity based on one of:
a result of the training procedure if the training procedure is caused to be executed, or
a result of the testing procedure if the testing procedure is caused to be executed.

2. The device of claim 1 , where the real transaction information is financial transaction information and the set of transaction values include at least one of:
a value indicating an identifier of an entity involved in a financial transaction,
a value indicating a transaction identifier,
a value indicating a transaction amount of the transaction amounts,
a value indicating a transaction date,
a value indicating an account balance,
a value indicating whether a transaction is a debit transaction or a credit transaction,
a value indicating a currency used in a transaction,
a value indicating a transaction type, or
some combination thereof.

3. The device of claim 1 , where the one or more transaction data types are associated with financial transactions and include at least one of:
a transaction amount data type,
an account balance data type,
a debit data type,
a credit data type,
a currency data type,
a transaction type data type, or
some combination thereof.

4. The device of claim 1 , where the one or more processors, when separating the set of transaction values into the one or more data streams, are to:
identify, for the set of transaction values, the one or more transaction data types, and
separate the set of transaction values into the one or more data streams,
where a transaction value of the set of transaction values is separated into a data stream, of the one or more data streams, based on a transaction data type of the one or more transaction data types.

5. The device of claim 1 , where the one or more processors, when training the model, are to:
train the model using one or more configurations parameters,
the one or more configuration parameters being associated with one or more sample statistical distributions that are intended to closely resemble the one or more actual statistical distributions.

6. The device of claim 5 , where the model is a Gaussian Mixture Model (GMM); and where the one or more processors, when training the GMM, are to:
determine a first sample statistical distribution of the one or more sample statistical distributions using the one or more configuration parameters,
determine that the first sample statistical distribution does not satisfy a threshold level of similarity with an actual statistical distribution of the one or more actual statistical distributions,
determine a second sample statistical distribution by modifying the one or more configuration parameters based on determining that the first sample statistical distribution does not satisfy the threshold level of similarity with the actual statistical distribution,
determine that the second sample statistical distribution satisfies the threshold level of similarity with the actual statistical distribution, and
identify the second sample statistical distribution as the sample statistical distribution to be used for the GMM.

7. The device of claim 1 , where the one or more processors, when generating the simulated transaction information, are to:
provide the random numbers as input for the model, and
generate the simulated transaction information using an inverse cumulative distribution function (ICDF).

8. A method, comprising:
receiving, by a device and for a set of transactions, real transaction information associated with a particular industry,
the real transaction information including a set of transaction values;

separating, by the device, the set of transaction values into one or more data streams based on one or more transaction data types,
the set of transaction values being associated with one or more actual statistical distributions,
a first set of transaction values, of the set of transaction values, being associated with transaction amounts,
the first set of transaction values being separated into a first data stream, and

a second set of transaction values, of the set of transaction values, being associated with a debit transaction or a credit transaction,
the second set of transaction values being separated into a second data stream;

training, by the device and using one or more machine learning techniques, a model for simulating the real transaction information;
generating, by the device, simulated transaction information by using random numbers as input for the model;
processing, by the device, the simulated transaction information to detect an anomaly relating to the real transaction information after generating the simulated transaction information;
causing, by the device, at least one of a training procedure or a testing procedure to be executed based on detecting the anomaly; and
providing, by the device, a recommendation to an entity based on one of:
a result of the training procedure if the training procedure is caused to be executed, or
a result of the testing procedure if the testing procedure is caused to be executed.

9. The method of claim 8 , where the real transaction information is financial transaction information and the set of transaction values include at least one of:
a value indicating an identifier of an entity involved in a financial transaction,
a value indicating a transaction identifier,
a value indicating a transaction amount of the transaction amounts,
a value indicating a transaction date,
a value indicating an account balance,
a value indicating whether a transaction is a debit transaction or a credit transaction,
a value indicating a currency used in a transaction,
a value indicating a transaction type, or
some combination thereof.

10. The method of claim 8 , where the one or more transaction data types are associated with financial transactions and include at least one of:
a transaction amount data type,
an account balance data type,
a debit data type,
a credit data type,
a currency data type,
a transaction type data type, or
some combination thereof.

11. The method of claim 8 , where separating the set of transaction values into the one or more data streams comprises:
identifying, for the set of transaction values, the one or more transaction data types, and
separating

Assignees

• Accenture Global Solutions Ltd

Inventors

• Han Jingguang
• Sirotkovic Jadran
• Wan Dadong
• Zuo Xin

Classifications

• G06N7/01

  Probabilistic graphical models, e.g. probabilistic networks · CPC title

• G06Q30/06 Primary

  Buying, selling or leasing transactions · CPC title

• G06Q20/22 Primary

  Payment schemes or models · CPC title

• G06N7/04

  Physical realisation · CPC title

• G06N7/06

  Simulation on general purpose computers · CPC title